AUDIT 7

The report recommends: "Segment A customers are low-value and acquisition spend should be reduced for this profile." Setting aside the modeling error, what is the most important business problem with this recommendation?

Segment A represents 54% of the customer base and averages 5.3 purchases per year. Labeling them as permanently low-value based solely on lower average income ignores segment volume, purchase frequency, and growth potential.

Based on the flaws identified in this report, which deployment decision is correct?

Do Not Deploy — the normalization error means the clusters are likely driven by income scale rather than real behavioral structure. The model must be re-run with normalized variables before any budget decisions are made.

The report chose k=2 because it "provides the clearest business distinction." Which analytical step is missing from this justification?

An elbow plot or silhouette score should be used to objectively evaluate whether k=2 fits the data better than k=3 or k=4 before committing to two clusters.

The submitted report states: "Variables were not normalized as they were already in their natural units." Income ranges from approximately $30,000 to $120,000; purchase frequency ranges from 2 to 12. What is the most accurate critique of this decision?

Because income has a range of roughly $90,000 while purchase frequency has a range of roughly 10, income will dominate the Euclidean distance calculation. The resulting clusters will mainly reflect income differences, not purchase patterns.